Yesterday at RSA, Google announced their new Cloud Security AI Workbench, a cybersecurity suite powered by an AI model coined Sec-PaLM that specializes in intelligent security LLM (Large Language Model).
Cloud Security AI Workbench will span a few new AI-powered tools (both owned by Google):
- Mandiant’s Threat Intelligence AI – finds, summarizes and acts on security threats
- VirusTotal – helps subscribers analyze and explain the behavior of malicious scripts
Google says that it plans to release the rest of the offerings to “trusted testers” soon, but in general, Sec-PaLM will assist customers in searching for and interpreting security events and interacting with the results conversationally. Also, users of Google’s Security Command Center AI will get explanations of attack exposures, assets effected, suggested mitigations and risk summaries, compliance and privacy findings.
Google’s play in the generative AI cybersecurity race comes just after Microsoft’s March release of competitor Security Copilot.
Jeffrey Sims, Principal Security Engineer, HYAS had this to say:
“Google has also been a large contributor to the open source LLM space with their fine-tuned model series called Flan (Fine-tuned LAnguage Net). These models range in parameter size (capability) and allow for commercial applications.
“In addition to Google’s offering, we’ll see many technologically advanced organizations leveraging these open source models which will allow for deep customization and creative use cases, working in tandem with AI Workbench’s “partner plug-in integrations,” mentioned above. The rate of innovation based on creative systems like this will radically accelerate the security space in the years to come. “
It will be interesting to see how Google’s offering competes against from companies like Orca Security and ARMO who are doing similar things. And it will be interesting to see who else jumps into this space.